Candidate Profile — Candidate #3
Position: Adobe — Applied Scientist - Multimodal (San Jose, CA; San Francisco, CA; Seattle, WA)
Candidate Location: San Francisco Bay Area
Experience: 9.6 years experience
Relevant Experience & Education Highlights
Expertise in multimodal generative AI, vision-language models, diffusion-based systems, and PyTorch aligns strongly with developing IP-aware guardrail mechanisms for Adobe Firefly's multimodal generative models.
1. Research & Technical Depth
PhD-level background in computer engineering combined with 9.6 years in applied ML and generative AI research provides deep technical foundation for the role.
- Published 10 papers and patents in CVML/GenAI conferences including CVPR, with 5 first-author publications selected as oral presentations (top 6/1000).
- Led multi-level taxonomy image segmentation model for product tagging at a major cloud computing company's Generative AI Innovation Center, integrating multimodal LLMs, multi-level segmentation, and Stable Diffusion for reimaging.
- Developed voice chatbot Q&A system using LLM-based multi-agent RAG knowledge base and network APIs for information retrieval.
- Co-led automated QA test case code generation with multi-modal graph-based LLMs and image feature extraction, achieving 80% effectiveness.
2. Vision-Language & Multimodal Reasoning
Hands-on experience with multimodal LLMs and foundation models supports advancing semantic IP understanding and real-time steering in generative systems.
- Integrated multimodal LLM classification with Stable Diffusion in fashion product tagging pipeline at a leading cloud provider.
- Applied computer vision techniques including image segmentation, object detection, and point clouds in perception systems for autonomous applications.
- Utilized PyTorch and TensorFlow for deep learning models in robotics perception, SLAM, and sensor fusion projects.
3. Rapid Scientific Experimentation
Proven ability to conduct rigorous experiments and achieve high-visibility results matches needs for evaluating trade-offs in creativity, fidelity, and IP safety.
- Achieved CEO-level visibility through 4 published papers on generative AI innovations at a major cloud services organization.
- Designed traffic light detection system for multi-lane intersections with 96% accuracy in L4 autonomous vehicle perception at an autonomous driving company.
- Led multi-sensor adverse weather detection system combining high-dimensional time-series data in 3D simulation platforms.
4. Inference-Time Alignment & Optimization
Experience optimizing multimodal pipelines and deploying large models enables low-latency production-scale inference without sacrificing quality.
- Deployed models using AWS SageMaker and cloud infrastructure for generative AI and perception systems.
- Developed GPU-accelerated workflows leveraging CUDA for deep learning and computer vision tasks.
- Optimized multi-sensor fusion including cameras, LiDARs, radars, GPS, and IMUs for real-time autonomous vehicle perception.
Requirements & Candidate Alignment
| Adobe Requirement | Candidate Qualification |
|---|---|
| Education: PhD or MS in Computer Science, Machine Learning, AI, or related field | PhD in Computer Engineering from a top-tier research university |
| Experience: 5+ years of experience in applied ML or generative AI research (industry or academia) | 9.6 years across generative AI innovation, autonomous perception, and robotics roles |
| Generative Models: Strong background in large-scale generative models (diffusion models, multimodal transformers, autoregressive systems) | Hands-on leadership with diffusion models including Stable Diffusion and multimodal transformers in product tagging pipelines |
| VLMs Expertise: Expertise in Vision-Language Models or multimodal foundation models | Integrated multimodal LLMs for classification, segmentation, and QA generation in generative AI projects |
| Frameworks: Proficiency in Python and modern ML frameworks (e.g., PyTorch), with experience of training and deploying large models | Proficient in Python, PyTorch, TensorFlow, Keras, Scikit-Learn, and AWS SageMaker for training and deployment |
| Experimental Skills: Strong experimental development and statistical evaluation skills | Led end-to-end experiments in multimodal systems, achieving 96% accuracy in detection and 80% in QA automation |
| Publications: Research contributions in controllable generation, alignment, AI safety, or multimodal learning. Publications in leading conferences (CVPR, ICCV, NeurIPS, ICML, ICLR, SIGGRAPH) or equivalent industry impact | 10 papers and patents with CVPR oral presentations and CEO-level impact in GenAI |
| Systems Rigor: Experience analyzing complex failure modes in multimodal systems. Understanding of large-scale inference systems and production ML constraints | Analyzed and optimized failure modes in multi-sensor perception and generative pipelines for production-scale deployment |
If you would like to discuss this candidate or other critical roles, here is a link to my calendar to schedule a call:
Schedule a CallContact:
Finding the signal in the noise since 2005